Software complexity and its impacts in embedded intelligent real-time systems

Applications of intelligent software systems are proliferating. As these systems proliferate, understanding and measuring their complexity becomes vital, especially in safety-critical environments. This paper proposes a model assessing the impacts of complexity for a particular type of intelligent software system, embedded intelligent real-time systems (EIRTS), and answers two research questions. (1) How should the complexity of embedded intelligent real-time systems be measured?, and (2) What are the impacts of differing levels of EIRTS complexity on software, operator and system performance when EIRTS are deployed in a safety-critical large-scale system? The model is tested and operationalized using an operational EIRTS in a safety-critical environment. The results suggest that users significantly prefer simple decision support and user interfaces, even when sophisticated user interfaces and complex decision support capabilities have been embedded in the system. These results have interesting implications for operators using complex EIRTS in safety-critical settings.